Issue with ColorChecker Profiles
I'm not able to get true blacks with these Profiles - the rest of the color seems to be in order, but the images look slightly less contrasty and no matter what I do, I can't get a true black (0 0 0 0) value with them.
With the regular camera profile it's no issue at all...
Happens on both Windows and OSX with current versions of Capture One and ColorChecker Camera Calibration...
Camera: Fujifilm GFX 50s
-
I see thank you
0 -
https://ninedegreesbelow.com/photography/well-behaved-camera-profile.html
@C-M-B use Argyll and or dcamprof and add true black and true white to the profile manually...
See section D5 at the included link...sorry tried t reply directly to you but can't see how on this forum..
0 -
I was struggling with the same issue of not getting 0,0,0 for blacks, but if I ensure the curve for the exported image is set to Linear Response instead of the default Auto, the black becomes 0,0,0 as expected. Seems to do the trick.
0 -
Even if you export the image to something like AdobeRGB or ProPhoto?
0 -
Yes, just confirmed. Run through:
1. When exporting for ColorChecker Passport make sure these are as follows: http://f.jere.co/EOxpuh
2. Imported photo: http://f.jere.co/Vq8avI
3. Pulled the curve to make things 0,0,0 (you can see this here in the Capture One too): http://f.jere.co/6yhoIQ
4. Exported with these settings: http://f.jere.co/9U4Lwc
5. And imported back to Capture One and it's still 0,0,0 (again you can see it on the top bar as well) http://f.jere.co/AH1GvGEDIT:
And indeed I was struggling with the very issue of not getting black anywhere near 0,0,0, but my mistake clearly was not setting the curve to Linear Response.0 -
Interesting. And you're sure you're using the calibrated profile? Because that would be a first.
0 -
Yes, triple checked. Capture One 21 Pro Build 14.0.2.49 and ColorChecker Passport 1.2.0 in use.
0 -
Just to be clear...
You're creating the profile with X-Rite software, correct?
Because, I always set C1 to Linear before exporting to create the profile, and I never achieved 0,0,0.
But, I'm not using C1 21, still on V.20.
Many thanks.
D.1 -
(I'm having the exact same experience as Denis)
0 -
One other thing...
In 3. in your post above, the screen in the link is blue, which is correct. The shadow warning is showing 0,0,0, for the entire image.
In 5. in your post, the screen in the link is black. I'm not clear what this means except that it's what I get when I try to achieve 0,0,0. In other words, the blue shadow warning fails to active with the profiled image set to zero. And, if you hover the mouse over the image when it's black, the RGB at the top of the screen reads something like 30,30,30, not the 0,0,0, you'd want.
Maybe I'm misunderstanding.
Just checking! :)
Thanks.
D.
0 -
I'm not using capture one. I use open source software either Rawtherapee or Darktable. I also have ON1 Photo which I really don't use that much. I follow the instructions for using Darktable and others much as you would capture 1 and feed it a linear 16 bit tiff or in ON1 the have a reference file mode to create the linear tiff. For some reason I get poor results. THe profiles are always creating dark unimpressive images. I have fed the xrite software source images of various exposures in case this was an issue and it seems to have little effect. Since xrite really doesn't say much about their software or profiles its hard to trouble shoot. I have essentially given up on it and use the open source tools Argyll or dcamprof to create profiles. At least there you can really get at the nuts and bolts of it. Although they are command line there are now nice GUI frontends to make them easier to use...one is called roughprofiler and the other Coca if anyone is interested http://www.dohm.com.au/coca/ and this one has mac and windows versions...http://www.jpereira.net/roughprofiler
0 -
One other comment, I know in DT you set the input color space and output/export to be equal (it doesn't matter what you select they just need to be the same) to preserve the color data so you create a "no op " with no colorspace conversions. In DT this is a pass through and the color is not modified. This is important as well as the linear tone curve to preserve the original color data. Capture one may be different in how it works???
0 -
@Denis Mortell. In 5 the shadow warning is not turned on, you can confirm this in the menu too.
0 -
This morning, I happened to open the X-Rite ICC Profiling software and I was prompted to update it. The new version is 2.2.0
I created a profile of an image of a passport CC and applied it to the image of the CC back in Capture One.
A few things......
1. When I zero out the image (e.g. by dragging the Levels shadow slider all the way to the right.) the blue shadow warning appears all over the image, as it should. With the old version of the X-Rite software, the profile made the image turn black, not blue, and the RGB reading was never lower than about 20,20,20.
2. However, hovering the colour picker over the image now gives a reading of about 1,1,1, not 0,0,0 as you would expect. But, it's very close.
3. Correcting for blue is the most critical colour when making a profile. After applying the profile to the file and increasing the exposure so that it regains its normal brightness (The green patch should be around 100, 150, 80, for this) the blue patch reading is superb.
Conclusion: They seem to have done something to improve it. I'd like a 0,0,0, reading. But, everything else is much better.
D.
Ps. After almost a year of Lumariver, I've almost given up. Hopeless Support. Avoid.
0 -
Version 2.2.0 has been out since September 2020, so it's not quite as new as one might think....
And yes, it's an improvement but still far from perfect and considering how long we've been waiting it's just no acceptable.
Regarding Lumariver...yeah I've heard that their support isn't great but the profiles are much better - that is to say if you're prepared to do all the legwork and really dig into profiling and you know which curves and settings will produce the desired profile.
Are you using the ColorChecker Passport or the ColorChecker Digital SG?
0 -
I have both charts.
The https://deltae.picturae.com/ results with this new version of X-Rite are better than Lumariver.
basICColor Profiler is probably the best, but it's very expensive.
D.
0 -
Just tried to create an ICC with another file - same camera, same lens.
When I applied the profile the CC back in C1, it didn't go shadow-warning blue when darkened. It went that old muddy black again. RBG minimum is about 7,7,7.
However, if I export a TIFF of this file and run it through DeltaE, it gets ****. Go figure, as they say.
I give up.
D.
0 -
I've been struggling with this for 3 years, hoping one day it will work.
The colour correction is great, but the elevated black levels are a killer.
As has been said before: as soon as you apply the icc profile created by the colourchecker calibration tool you lose true blacks and contrast, no matter what you do to the image sliders.
I tried it on a totally black zero exposure image and as soon as I apply the xrite the levels pop up from 0,0,0 to 0,3,3
wish there was a solution
1 -
For me the solution was ditching the xRite stuff, using an IT8.7 target and a free software called CoCa to get ICC profiles.
The software has been discontinued for a while but the results are pretty good (considering the amount of colour patches it could be better than the ColourChecker) and it's able to get true black tones.
0 -
You are really using Argyllcms which is open source and readily available. CoCo is just a GUI for this command line app. It is very nice and easy. There are a couple of others too that use Argyll.
There is a nice worked example here of the commandline stuff in the background.....esp sections D5 and D6 where you can add perfect black and white to your profile to make it better behaved.. https://ninedegreesbelow.com/photography/well-behaved-camera-profile.html#step-by-step-procedure
...
0 -
Exactly, and I have to say it's a bit shocking to see how good the results can be considering the low cost of the whole thing. All you need is the IT8-Target and proper lighting technique - the rest is very easy to do and I think CaptureOne could (and should) easily implement it in its software.
0 -
I use Darktable which recently introduce a nice feature to enhance color calibration. After the image is WB and has the input profile applied you can use an image of a colorchecker and it will do an optimization and create a set of channel mixer coefficients..essentially and adjusted matrix to further tweak and improve color accuracy...Its quick and gives you a deltaE report....no tiff file creation or any other messing around so even a less than perfect profile can be tweaked and saved as a preset to apply to a group of similar images....It would be a nice feature in C1 https://darktable-org.github.io/dtdocs/module-reference/processing-modules/color-calibration/#extracting-settings-using-a-color-checker
0 -
I just asked this in another thread but realised it had been marked as 'complete' so am not sure if I'll get much action with that one..
I've read all of these posts now, twice even, and am none the wiser as to what an actual workflow for colour correction looks like. What is a definitive answer after all of these posts?
Can we just hammer support to change this (even though it has been over 12 years since people have asked for this x-rite to be implemented into C1..)
Is it even worth buying a colour checker given this issue? Or are there alternatives out there which would do the job instead? Thanks so much
1 -
Forget the ColorChecker. Really forget it. It does not work.
The only thing that works for me is using Coca (which is basically a GUI for Agryll) with an IT8.7 target and those results are fantastic. True to life colours, true to life contrast, black is black and white is white.
My workflow:
Take image (as RAW with ICC set to "no color correction" and curve set to "Linear Response") making sure that the white patch is approx. 245/245/245 (RGB), select a grey patch for white balance and try to avoid any reflections shadows other issues that could interfere with the target.
Then export as 16bit TIFF with ICC set to "embed camera profile".Next I load the image into CoCa, select IT8.7 as target and I use the reference file from the target manufacterer. Export using Gamma + matrix, Relative Colorimetric at Ultra High Quality.
The only tweak I apply is a white point adjustment of 1,20 which get me a slightly(!) softer graduation in the highlights - but other than that it's perfect.Then you save the ICC profile in your Capture One Color Profile folder using the Camera name (e.g. FujiGFX50S") followed by some kind of designation like "Calibrated".
Then you reload CaptureOne and you can select the ICC from the drop-down menu. Remember to also set the curve to "Film Standard", "Film High Contrast" or "Film Extra Shadow". If you select any of the custom curves it'll default back to the generic ICC profile without even telling you.
1 -
Do you tweak the gamma setting??
0 -
Not at all, it's set to 2.2 which is the standard setting and most widely used.
0 -
C-M-B's post is a reference post and, more or less in a nutshell, what I ended up doing after weeks of messing around.
I accept I might be misunderstanding something, but after days/weeks of trying and reading these posts, I couldn't get the results I needed using X-rite Color Checker for C1.
As per C-M-B post: The 1st thing is to ditch the x-rite ColorChecker camera calibration software for creating an icc profile to use in C1. Do this, and you will save hours/days/weeks of tinkering and 2nd guessing. It does give good colour accuracy, but unless set to proof view of export with embedded camera profile, the black point behaviour is odd. And, I export for web and work in RAW then proof using sRGB embedded icc profile.
Following advice on this forum, I found CoCa and that immediately gave me accurate colours without losing the clarity from the odd black point x-rite software creates.
But initially, CoCa gave an odd curve to WP, which was great for helping to create a white background (product photography shots) but too punchy and not accurate and lost highlight detail. I used 1.3 to help and perceptual (as it was the default and didn't fully understand) but wasnt entirely happy. But now I use 1.2 and Relative Colorrimtric as per C-M-B post, and I think it's spot on. Also, I use a Linear response and gently recover some punch to the image elsewhere.
Ref: Buying the color checker.I use the x-rite color check passport; it cost about £100, and I delayed purchasing because of cost, but I now consider it one of my best purchases. And, I wish I had bought it when I first started. Without a genuine industry-standard patch (bought from a reputable supplier) then stick to the generic icc in C1, which is still a very good option.
For CoCa you can use the colour checker passport. It has a Color Checker Passport as one of the options/reference files. But I found it easier to crop to the 24 patches and use X rite ColorChecker 24 patch options within CoCa. I didnt want the work involved transposing all the extra x-rite patch data needed for the passport into a .cie file CoCa could use - more on this later.
Here is the important thing:
Colour Checker (both 24 patch and passport) changed their inks in 2014 but CoCa uses genric reference files averaged from results that I think relate to the old patch ink.
You can download the newer after 2014 new X-rite reference files here:
https://xritephoto.com/ph_product_overview.aspx?ID=1192&Action=Support&SupportID=5884&catid=28But they dont work in CoCa as the panel mapping and some other things are different. Here is the some expert info regarding this:
https://tomassobekphotography.co.nz/articles/create-CIE-file-from-spectral-ColorChecker-Passport.phpAnd another essential site for reference:
https://www.babelcolor.com/colorchecker.htmYou could use the Toma Sobek 24 patch .cie file here as this works in CoCa and should be very accurate: https://tomassobekphotography.co.nz/articles/1/myColorChecker.cie
However, following the info mostly from Tomas Sobek, I worked out the patch mapping, then hand-copied the new post 2014 X-rite reference data into the format that CoCa accepted for a custom reference file for the X-rite ColorChecker 24 patch.
Also if using a >2014 x-rite 24 patch then expose the white patch to 241 it used to be 245 before 2014 but now it's 241.
I use my own custom .cie file I created using the X-rite after 2014 reference data from their website and it has worked perfectly for me and I have super accurate colors under the lighting conditions I created the profile. Until there is enough data to form a new average I think its best to use the x-rite reference after 2014 data.
I hope I've done everything right, but please let me know if I've got stuff wrong.
Here is the text content from my file using the X-rite after 2014 reference data. You need to cut and paste the text below into a text file and then save, then rename the file type as .cie (replace .txt) it then works within CoCa:
IT8.7/2
ORIGINATOR "Michael Gates X-Rite ColorChecker24_AFTER_Nov2014 reference data"
DESCRIPTOR "ColorChecker 24"
CREATED "10/9/21 using x-rite data from 4/28/2015 # Time: 14:33 i1Pro 2 ; Serial number 1001785"
MANUFACTURER "X-Rite/Gretag Macbeth"NUMBER_OF_FIELDS 4
BEGIN_DATA_FORMAT
SAMPLE_ID LAB_L LAB_A LAB_B
END_DATA_FORMATNUMBER_OF_SETS 24
BEGIN_DATA
A01 37.54 14.37 14.92
A02 64.66 19.27 17.5
A03 49.32 -3.82 -22.54
A04 43.46 -12.74 22.72
A05 54.94 9.61 -24.79
A06 70.48 -32.26 -0.37
B01 62.73 35.83 56.5
B02 39.43 10.75 -45.17
B03 50.57 48.64 16.67
B04 30.1 22.54 -20.87
B05 71.77 -24.13 58.19
B06 71.51 18.24 67.37
C01 28.37 15.42 -49.8
C02 54.38 -39.72 32.27
C03 42.43 51.05 28.62
C04 81.8 2.67 80.41
C05 50.63 51.28 -14.12
C06 49.57 -29.71 -28.32
D01 95.19 -1.03 2.93
D02 81.29 -0.57 0.44
D03 66.89 -0.75 -0.06
D04 50.76 -0.13 0.14
D05 35.63 -0.46 -0.48
D06 20.64 0.07 -0.46
END_DATA0 -
Just to add, the Pro Colour feature of Capture 21 is a big fib.
While it might be more accurate than the generic profile it is NOT good enough for someone using it professionally for product photography and should NOT be named Pro. I'd even say on my 5D2 (this oldy is still good enough and I love it) it is less accurate and too warm in the warmer hues.
It is nowhere near as close as the X-rite (even with its dodgy black level) custom icc and nowhere near as good as the CoCa generated profile.
Why invest so much effort into this?????? It's pointless, it's just slightly different than the generic, and brings out colours in a sunset but is NOT more accurate.
Why not, instead, invest time for proper support for generating a working colour profile with x-rite. This is something people using capture one for customers and to make money DO use and that WOULD be Pro level.
0 -
c1 has cultural heritage produkt line with bigger interests in there then just software and much more expencive. why should they give you as normal consumer with standart c1 version something that they sell for much more?..
0 -
Maybe because their competition offers it for much less?...
0
Post is closed for comments.
Comments
101 comments